A method for intelligent evaluation of instant green tea quality based on multi-source data fusion and application thereof

Abstract

The present application relates to a kind of instant green tea quality intelligent evaluation method and application based on multi-source data fusion, belong to food detection field, is related to the system construction of instant green tea quality intelligent identification.The present application mainly provides a kind of fast tea leaf intelligent quality evaluation technology, the quality index of instant green tea is determined by chemical analysis method, and relevant data are collected using intelligent evaluation technology and the relationship between quality index and intelligent characteristics is revealed using correlation analysis.Based on independent data source, instant green tea variety classification model is constructed.The present application combines multi-source data fusion technology and machine learning algorithm, and constructs a set of efficient, accurate, comprehensive instant green tea quality intelligent evaluation system and method, not only realizes the classification of instant green tea variety and the prediction of quality index, but also compares the fusion strategy between different quality index data, the model performance is optimized by introducing feature selection method, to further improve the practicability and reliability of system.

Description

Technical Field

[0001] This invention relates to the field of food testing technology, and in particular to an intelligent evaluation method and application for the quality of instant green tea based on multi-source data fusion. Background Technology

[0002] Instant green tea, as a convenient and quick beverage, is widely popular due to its unique flavor and health benefits. With the accelerating pace of modern life, consumers' demands for the quality of instant tea are constantly increasing. However, the quality differences among different varieties of instant green tea significantly influence consumer choices and brand market competitiveness. Currently, traditional quality evaluation methods typically rely on human sensory evaluation and chemical analysis, but their subjectivity and inconsistency limit the accuracy and reliability of the assessment. Therefore, there is an urgent need to seek more objective and efficient intelligent evaluation technologies to improve the accuracy and consistency of instant green tea quality testing. With the rapid development of intelligent evaluation technologies, it has become possible to evaluate the quality of instant green tea using computer vision, electronic noses, and electronic tongues. Meanwhile, the content of important compounds in instant green tea, such as tea polyphenols, catechins, and caffeine, directly affects the quality and nutritional value of instant green tea. Therefore, researching rapid quantitative analysis methods for key components not only has theoretical significance but also practical application value.

[0003] In recent years, the rapid development of intelligent evaluation technologies has provided new perspectives and methods for tea quality assessment. These technologies include advanced sensor technologies such as computer vision, electronic nose, and electronic tongue, which can efficiently and accurately conduct comprehensive analysis of the appearance, aroma, and taste of tea.

[0004] Currently, research on the application of intelligent evaluation technology in the quality testing of instant green tea is still relatively scarce both domestically and internationally. Existing research mainly focuses on quality testing aspects such as tea variety differentiation, grade differentiation, geographical origin identification, and processing monitoring.

[0005] Although there is a lack of systematic data in the application research of instant green tea, it is hoped that by introducing intelligent evaluation technology and combining it with machine learning algorithms, a more accurate and rapid detection and analysis of the quality of instant green tea can be achieved. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the existing technology by providing an intelligent evaluation method and application for the quality of instant green tea based on multi-source data fusion. Through advanced technical means, this method achieves efficient, accurate and comprehensive evaluation of the quality of instant green tea, thereby overcoming the limitations of traditional quality evaluation methods, such as high cost, complex operation and susceptibility to subjective factors.

[0007] This invention can be achieved through the following technical solutions:

[0008] The first objective of this invention is to provide an intelligent evaluation method for the quality of instant green tea based on multi-source data fusion, the intelligent evaluation method comprising the following steps:

[0009] Step 1: Select instant green tea as raw material. Brew a portion of the instant green tea with boiling water to make tea soup, and use the other portion of instant green tea as tea powder. Use a computer vision device to collect visual images of the tea soup and tea powder, use an electronic nose to obtain aroma information of the tea soup and tea powder, use an electronic tongue to collect taste information of the tea soup, and use high performance liquid chromatography to measure quality component information in the tea soup and tea powder.

[0010] Step 2: The aroma, visual and taste information are used as feature parameter data for dimensionality reduction. After dimensionality reduction, the dataset is divided, the quality indicators of tea soup and tea powder are measured, and correlation and significance analysis are performed on the quality indicators and feature parameters to form a fused feature set.

[0011] Step 3: Perform feature selection on the fused feature set formed in Step 2 to filter out the optimal feature subset. Based on the optimal feature subset, construct SVM, KNN, and RF regression classification models and PLSR, SVR, and RF as prediction models to predict the content of quality indicators of instant green tea. Use classification accuracy, R² of the training set and prediction set as the benchmarks. 2 The evaluation criteria include indicators such as value, RMSEC, RMSEP, and RPD.

[0012] Step Four: Based on the optimal feature subset formed in Step Three, decision-level and feature-level data fusion were applied. The decision-level model architecture employed a multiple linear regression (MLR) model and DS evidence theory. At the feature level, feature selection methods were introduced to perform multi-source data fusion. Based on the evaluation results, a fusion model was obtained to improve the accuracy and stability of classification and prediction. The decision-level model architecture was used for identifying the variety of instant green tea, while the feature-level model architecture was used for identifying the quality of instant green tea. R... 2 Values, RMSEC, RMSEP, and RPD are used as evaluation metrics to select the best fusion model.

[0013] Furthermore, the computer vision device includes:

[0014] Darkroom, tripod, light source, camera, and container;

[0015] The bracket, light source, camera, and container are all located inside the dark box;

[0016] The light source and camera are respectively connected to the bracket;

[0017] The camera, light source, and container are arranged in order from top to bottom;

[0018] The container is used to hold tea soup or tea powder.

[0019] Furthermore, the darkroom includes a black matte acrylic panel in the shape of a cube (400 mm × 400 mm × 600 mm). A side sliding groove and a bottom groove are provided on the front side of the darkroom to allow the front panel to slide up and down and be inserted into the bottom groove, so that the interior of the darkroom forms a closed space to avoid the external light environment from affecting the sample collection.

[0020] The support includes a support platform (300 mm × 300 mm), a support rod (400 mm high), and an adjustment frame. The support platform and the support rod are connected, and the adjustment frame is connected to the support rod. A stepped circular groove is designed at the center point of the support platform to fix the container and ensure that the collection position is consistent each time. The adjustment frame is used to adjust the distance between the light source and the camera and the container.

[0021] The light source adopts the D65 ring shadowless light source specified by the International Commission on Illumination (CIE), which is close to the true resolution of daylight. Its color temperature is 6500 K, inner diameter is 75 mm and outer diameter is 120 mm. It can emit uniform light and thus obtain higher quality digital image information.

[0022] The camera used is a Nikon D5100 (CMOS), which is connected to a computer via a data cable and the camera parameters are adjusted using software to achieve image acquisition.

[0023] Furthermore, the olfactory sensing detection system includes an electronic nose.

[0024] Furthermore, the taste sensing system includes an electronic tongue.

[0025] Furthermore, the regression classification model (discriminative model) includes one or more of SVM, KNN, and RF;

[0026] The prediction model includes one or more of PLSR, SVR, and RF.

[0027] Furthermore, the model evaluation indicators and calculation methods in the evaluation criteria are as follows:

[0028] Classification accuracy ;

[0029] Coefficient of determination R 2 value ;

[0030] Cross-validation root mean square error ;

[0031] Prediction root mean square error ;

[0032] Standard deviation ;

[0033] relative percentage deviation .

[0034] In the formula, n is the number of samples in the dataset; X i This refers to the actual value of the i-th sample during the prediction model building process; Y is the average of the actual values ​​of all samples during the prediction model building process; i The predicted value for the i-th sample during the model building process; This is the average of the predicted values ​​for all samples during the model building process.

[0035] Furthermore, the determination of the quality indicators of the tea infusion and tea powder materials includes the following process:

[0036] The components in tea leaves, including TP (tea polyphenols), FAA (total free amino acids), TP / FAA (ratio of tea polyphenols to total free amino acids), CAF (caffeine), EGC (epigallocatechin gallate), C (catechins), EC (epicatechins), and EGCG (epigallocatechin gallate), were determined by liquid chromatography. For tea powder materials: the tea sample was ground into powder, homogenized, and then ultrasonically extracted using ultrapure water or a specific solvent. The extract was centrifuged, and the supernatant was filtered before analysis. A C-18 reversed-phase column was used to determine the contents of tea polyphenols, free amino acids, caffeine, and catechins according to national standards, and the data were recorded.

[0037] Furthermore, in step two, PCA dimensionality reduction is used.

[0038] Furthermore, the feature selection method includes one or more of the following four feature selection methods: Pearson score, recursive feature elimination (RFE), particle swarm optimization (PSO), and Lasso regression.

[0039] Furthermore, Pearson correlation analysis was performed on nine color characteristic parameters and quality indicators of instant green tea infusion and tea powder. Simultaneously, 14 aroma characteristic parameters of tea infusion and tea powder were extracted using electronic nose technology, and the correlation between these aroma characteristics and 10 quality indicators was analyzed in detail. In addition, correlation analysis was conducted on 18 taste characteristic parameters extracted by an electronic tongue sensor and 10 quality indicators of instant green tea. Specifically, TP content was determined according to GB / T 8313-2018 "Determination Method of Tea Polyphenols and Catechins in Tea". Catechins and CAF content were determined according to GB / T 8313-2018 "Determination Method of Tea Polyphenols and Catechins in Tea". FAA content was determined according to GB / T 8314-2013 "Determination of Total Free Amino Acids in Tea".

[0040] Furthermore, in the decision-level data fusion, a multiple linear regression (MLR) model was used to classify different varieties of instant green tea, and the quality of green tea was evaluated using DS evidence theory. Simultaneously, Pearson regression, recursive feature elimination (RFE), particle swarm optimization (PSO), and Lasso regression methods were used to evaluate the R-values ​​of the six models. 2 The values, root mean square error correction set (RMSEC), root mean square error prediction set (RMSEP), and relative prediction bias (RPD) are evaluated to select the best fusion model.

[0041] The second objective of this invention is to provide an application of an intelligent evaluation method for the quality of instant green tea based on multi-source data fusion, which is used for intelligent identification and rapid prediction of the quality of instant green tea.

[0042] Compared with the prior art, the present invention has the following advantages:

[0043] 1. This invention provides a method and application for intelligent quality evaluation of instant green tea based on multi-source data fusion. The core of this method lies in the organic combination of three intelligent evaluation technologies: computer vision, electronic nose, and electronic tongue, constructing a multi-source data fusion platform. Computer vision technology, through image acquisition and processing, can accurately extract the color features of instant green tea powder and tea liquor. These features intuitively reflect key information such as the oxidation degree and pigment distribution of instant green tea. Electronic nose technology simulates the olfactory system of mammals, enabling objective evaluation of the aroma quality of tea by detecting the aroma components of instant green tea. Electronic tongue technology further simulates the human taste system, accurately analyzing the taste characteristics of instant green tea and providing more comprehensive data support for quality evaluation.

[0044] 2. This invention provides an intelligent evaluation method and application for the quality of instant green tea based on multi-source data fusion. Based on data collection, this invention utilizes various machine learning algorithms, including Support Vector Machine (SVM), Random Forest (RF), and K-Nearest Neighbors (KNN), to construct instant green tea variety classification models and quality index prediction models. Through training and learning on a large amount of experimental data, these models can accurately identify different varieties of instant green tea and predict the content of key quality indicators, such as tea polyphenols, catechins, and caffeine. Experimental results show that the machine learning-based models exhibit excellent performance in both classification and prediction tasks, significantly outperforming traditional chemical analysis and sensory evaluation methods.

[0045] 3. This invention provides a method and application for intelligent quality evaluation of instant green tea based on multi-source data fusion. To further improve the accuracy and generalization ability of the model, this invention also introduces feature-level fusion and decision-level fusion strategies. The feature-level fusion strategy integrates data from different intelligent evaluation technologies to form a more comprehensive quality evaluation dataset. The decision-level fusion strategy constructs multiple classification or prediction models based on independent data sources and integrates the decision results of these models through certain fusion rules to arrive at the final judgment. Experiments show that the application of the fusion strategy significantly improves the classification accuracy and prediction precision of the model.

[0046] 4. This invention provides an intelligent evaluation method and application for the quality of instant green tea based on multi-source data fusion. In terms of model optimization, this invention introduces various feature selection methods such as Particle Swarm Optimization (PSO), Recursive Feature Elimination (RFE), and Lasso Regression. These methods reduce data redundancy and improve the model's prediction accuracy and generalization ability by optimizing feature subsets. Experimental results show that the model optimized using feature selection methods exhibits higher accuracy and stability in both classification and prediction tasks.

[0047] In summary, this invention, by combining multi-source data fusion technology and machine learning algorithms, constructs a highly efficient, accurate, and comprehensive intelligent evaluation system and method for the quality of instant green tea. This system enables the classification of instant green tea varieties and the prediction of quality indicators, providing strong support for the production, quality control, and marketing of instant green tea. Furthermore, by introducing feature selection methods to optimize model performance, the system's practicality and reliability are further enhanced, paving a new path for the intelligent development of the tea industry. Attached Figure Description

[0048] Figure 1 In an embodiment of the present invention, images of tea powder and tea soup are identified using computer vision, wherein (A) is tea powder and (B) is tea soup.

[0049] Figure 2In an embodiment of the present invention, the electronic nose of different types of instant green tea showed the results of (A) PCA and (B) LDA of tea powder, (C) PCA and (D) LDA of tea infusion.

[0050] Figure 3 In an embodiment of the present invention, the electronic tongue (A) PCA results and (B) LDA results of different types of instant green tea are presented.

[0051] Figure 4 In the embodiments of the present invention, the catechin content of different types of instant green tea is (A) EGC (B) C (C) EC (D) EGCG (E) ECG (F) TC. Note: Different lowercase letters indicate significant differences. p < 0.05).

[0052] Figure 5 This is a schematic diagram of the classification and fusion strategy of the intelligent evaluation method for the quality of instant green tea based on multi-source data fusion of the present invention, (A) feature-level fusion, (B) decision-level fusion.

[0053] Figure 6 This is a flowchart illustrating the intelligent quality evaluation method for instant green tea based on multi-source data fusion according to the present invention. Detailed Implementation

[0054] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0055] Unless otherwise specified in this technical solution, the component model, material name, connection structure, application, algorithm, and other features are considered to be common technical features disclosed in the prior art.

[0056] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Terms such as "upper," "lower," "left," and "right" are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may change. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0057] It should be noted that in this invention, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0058] In the following embodiments, unless otherwise specified, the raw materials or processing techniques are all commercially available materials or conventional processing techniques in the art.

[0059] This invention relates to a method and application for intelligent quality evaluation of instant green tea based on multi-source data fusion, belonging to the field of food testing, and involves the construction of a system for intelligent identification of instant green tea quality. This invention mainly provides a rapid intelligent quality evaluation technology for tea. It determines the quality indicators of instant green tea through chemical analysis and collects relevant data using intelligent evaluation technology. Then, correlation analysis is used to reveal the relationship between quality indicators and intelligent features. Based on independent data sources, a classification model for instant green tea varieties is constructed. This invention, by combining multi-source data fusion technology and machine learning algorithms, constructs an efficient, accurate, and comprehensive intelligent quality evaluation system and method for instant green tea, realizing the classification of instant green tea varieties and the prediction of quality indicators. By introducing feature selection methods to optimize model performance, the practicality and reliability of the system are further improved.

[0060] In the following embodiments:

[0061] The computer vision device includes: a dark box, a support, a light source, a camera, and a container; the support, light source, camera, and container are all located inside the dark box; the light source and camera are respectively connected to the support; the camera, light source, and container are arranged sequentially from top to bottom; the container is used to hold tea infusion or tea powder. The dark box includes a black matte acrylic panel, shaped like a cube, with a side sliding groove and a bottom groove on the front side of the dark box to allow the front panel to slide up and down and insert into the bottom groove, forming a closed space inside the dark box to avoid the influence of external light environment on sample acquisition; the support includes a platform, a support rod, and an adjustment frame; the platform and support rod are connected, and the adjustment frame is connected to the support rod; a stepped circular groove is designed at the center point of the platform to fix the container, ensuring consistent acquisition position each time, and the adjustment frame is used to adjust the distance between the light source, camera, and container; the light source is a ring-shaped shadowless light source; the camera is connected to a computer via a data cable and its parameters are adjusted using software to achieve image acquisition.

[0062] The dark box includes a black matte acrylic panel in the shape of a cube (400 mm × 400 mm × 600 mm). A side sliding groove and a bottom groove are provided on the front side of the dark box to allow the front panel to slide up and down and be inserted into the bottom groove, so that the interior of the dark box forms a closed space to avoid the external light environment from affecting the sample collection.

[0063] The support includes a support platform (300 mm × 300 mm), a support rod (400 mm high), and an adjustment frame. The support platform and the support rod are connected, and the adjustment frame is connected to the support rod. A stepped circular groove is designed at the center point of the support platform to fix the container and ensure that the collection position is consistent each time. The adjustment frame is used to adjust the distance between the light source and the camera and the container.

[0064] The light source adopts the D65 ring shadowless light source specified by the International Commission on Illumination (CIE), which is close to the true resolution of daylight. Its color temperature is 6500 K, inner diameter is 75 mm and outer diameter is 120 mm. It can emit uniform light and thus obtain higher quality digital image information.

[0065] The camera used is a Nikon D5100 (CMOS), which is connected to a computer via a data cable and the camera parameters are adjusted using software to achieve image acquisition.

[0066] The electronic nose uses the CNose-14 electronic nose manufactured by Shanghai Baosheng Industrial Development Co., Ltd., and consists of a sample introduction system, a sensor array, a signal acquisition system, and a gas cleaning channel. The sample introduction system is responsible for collecting sample gas and controlling the gas flow rate; the sensor array, the core component of the electronic nose system for detecting sample gas, consists of 14 imported metal-oxide-semiconductor (MOS) sensors with different properties, exhibiting varying sensitivities to different volatile compounds; the signal acquisition system controls the electronic nose system to acquire response signals.

[0067] Flavor acquisition using the electronic tongue employs the Smart Tongue electronic tongue manufactured by Shanghai Baosheng Industrial Development Co., Ltd. It consists of a sample introduction system, a sensor array, and a data acquisition and analysis system. The sensor array comprises six cross-sensitive chemical sensors (platinum S1, gold S2, palladium S3, tungsten S4, titanium S5, and silver S6). Before each test, the electronic tongue system requires preheating for at least 30 minutes, and the sensors are cleaned with deionized water. It is also ensured that the protective solution inside the Ag / AgCl reference electrode has not dried out. During sample detection, the sample in the beaker is immersed in the bottom of the electrode, causing the taste signal to induce a change in the potential of the lipid-sensitive membrane through the sensor array, which is then converted into an electrical signal. Finally, the software receives, records, processes, and saves the electrical signal response values ​​to a computer.

[0068] Example

[0069] like Figures 5-6 As shown in the figure, this embodiment provides a method and application for intelligent evaluation of the quality of instant green tea based on multi-source data fusion. The intelligent evaluation method includes the following steps:

[0070] Step 1: Select instant green tea as raw material. Brew a portion of the instant green tea with boiling water to make tea soup, and use the other portion of instant green tea as tea powder. Use a computer vision device to collect visual images of the tea soup and tea powder, use an electronic nose to obtain aroma information of the tea soup and tea powder, use an electronic tongue to collect taste information of the tea soup, and use high performance liquid chromatography to measure quality component information in the tea soup and tea powder.

[0071] Step 2: The aroma, visual and taste information are used as feature parameter data for dimensionality reduction. After dimensionality reduction, the dataset is divided, the quality indicators of tea soup and tea powder are measured, and the correlation analysis between the quality indicators and feature parameters is performed to form a fused feature set.

[0072] Step 3: Perform feature selection on the fused feature set formed in Step 2 to filter out the optimal feature subset. Based on the optimal feature subset, construct SVM, KNN, and RF regression classification models and PLSR, SVR, and RF as prediction models to predict the content of quality indicators of instant green tea. Use classification accuracy, R² of the training set and prediction set as the benchmarks.2 The evaluation criteria include indicators such as value, RMSEC, RMSEP, and RPD.

[0073] Step Four: Based on the optimal feature subset formed in Step Three, decision-level and feature-level data fusion were applied. The decision-level model architecture employed a multiple linear regression (MLR) model and DS evidence theory. At the feature level, feature selection methods were introduced to perform multi-source data fusion. Based on the evaluation results, a fusion model was obtained to improve the accuracy and stability of classification and prediction. The decision-level model architecture was used for identifying the variety of instant green tea, while the feature-level model architecture was used for identifying the quality of instant green tea. R... 2 The fusion model is evaluated using metrics such as RMSEC, RMSEP, and RPD to select the optimal fusion model. The model used in step four is primarily based on statistical methods for feature fusion, employing different fusion strategies such as decision-level and feature-level fusion. Step three utilizes a regression classification prediction model, following a progressive approach of prediction followed by feature association analysis.

[0074] Specifically, the steps include the following:

[0075] Six types of instant green tea (purchased from Zhejiang Mingbao Company): G302, G305, G306, G307, J753, and ZTHG505 were selected and tested using computer vision, electronic nose, and electronic tongue, respectively, in the form of tea powder and tea infusion.

[0076] In computer vision inspection, a self-built computer vision device was used to acquire images of tea powder and tea infusion samples. 3g of tea powder was weighed and evenly spread in a glass dish. For the tea infusion sample, 0.5g of tea powder was brewed with 150mL of ultrapure water (boiling water), and the same amount was then placed in a porcelain tasting cup. All images were acquired under the same conditions and saved as JPG files. Subsequently, Python software was used to preprocess the images, automatically selecting a 600×600 pixel region of interest near the center point and extracting red, green, and blue channels, as well as multiple color feature parameters such as hue, saturation, and brightness. Each sample was prepared six times, with each iteration performed five times, resulting in 360 sets of data. The images of tea powder and tea infusion identified using computer vision are shown below. Figure 1 As shown, where Figure 1 (A) is tea powder. Figure 1 (B) is tea infusion. A preliminary experiment was conducted to determine the optimal detection parameters before electronic nose detection. The tea powder sample volume was 5g, the headspace time was 60min, and the ambient temperature was 25℃; the tea infusion sample volume was 15mL, with other conditions remaining the same. After the samples were sealed and allowed to stand in headspace vials, the sample gas was aspirated at a specific rate for detection. Each sample was also tested 6 times, with 5 tests per test, resulting in a total of 360 datasets. Figure 2The diagram shows the results of using an electronic nose to test different types of instant green tea. Figure 2 (A) Results of PCA on tea powder and Figure 2 (B) Results of LDA analysis of tea powder Figure 2 (C) shows the PCA results of the tea infusion and Figure 2 (D) represents the LDA result of the tea infusion.

[0077] For electronic tongue detection, the optimal detection parameters for tea infusion samples were determined based on preliminary experimental results. 0.5g of tea powder was dissolved in 150mL of ultrapure water, and 30mL was used as the test sample. After setting the acquisition frequency, sensor signal amplification factor, and detection time, the sample was immersed in the bottom of the electrode for detection, and the data was exported for analysis and processing. Each sample was tested six times, with five tests per test, resulting in a total of 180 datasets. Figure 3 The diagram shows the results of electronic tongue analysis for different types of instant green tea. Figure 3 (A) Results of PCA of tea infusion and Figure 3 (B) shows the LDA results of the tea infusion.

[0078] The determination of the quality indicators of the tea infusion and tea powder materials includes the following process:

[0079] The components of tea leaves, including TP (tea polyphenols), FAA (total free amino acids), TP / FAA (ratio of tea polyphenols to total free amino acids), CAF (caffeine), EGC (epigallocatechin gallate), C (catechins), EC (epicatechins), and EGCG (epigallocatechin gallate), were determined by liquid chromatography. Tea samples were ground into powder, homogenized, and then subjected to ultrasonic extraction with ultrapure water or a specific solvent. The extract was centrifuged, and the supernatant was filtered before analysis. A C-18 reversed-phase column was used to determine the contents of tea polyphenols, free amino acids, caffeine, and catechins according to national standards, and the data were recorded.

[0080] Pearson correlation analysis was performed on nine color characteristic parameters and quality indicators of instant green tea infusion and tea powder. Simultaneously, 14 aroma characteristic parameters of tea infusion and tea powder were extracted using electronic nose technology, and the correlation between these aroma characteristics and 10 quality indicators was analyzed in detail. Furthermore, correlation analysis was conducted on 18 taste characteristic parameters extracted by an electronic tongue sensor and 10 quality indicators of instant green tea. Specifically, TP content was determined according to GB / T 8313-2018 "Determination Method of Tea Polyphenols and Catechins in Tea". Catechins and CAF contents were determined according to GB / T 8313-2018 "Determination Method of Tea Polyphenols and Catechins in Tea". FAA content was determined according to GB / T 8314-2013 "Determination of Total Free Amino Acids in Tea".

[0081] In the modeling process, two methods were employed: classification models and regression models. The classification models (regression classification models) included Support Vector Machines (SVM), K-Nearest Neighbors (KNN), and Random Forests (RF) for qualitative identification of instant green tea varieties; the prediction models included Partial Least Squares Regression (PLSR), Support Vector Regression (SVR), and Random Forest Regression for rapid quantitative prediction of quality indicators. For example... Figure 4 The figure shows the content of catechins EGC, C, EC, EGCG, ECG, and TC in different types of instant green tea.

[0082] The model evaluation indicators and calculation methods in the evaluation criteria are as follows:

[0083] Classification accuracy ;

[0084] Coefficient of determination R 2 value ;

[0085] Cross-validation root mean square error ;

[0086] Prediction root mean square error ;

[0087] Standard deviation ;

[0088] relative percentage deviation .

[0089] The dataset was z-score normalized before modeling to eliminate the influence of unit dimensions. The data was divided into a 70% training set and a 30% test set, and 10-fold cross-validation was used to evaluate the model performance. Specifically, the original dataset was evenly divided into 10 mutually exclusive subsets, and each time 9 subsets were used for training and 1 subset for testing, repeated 10 times. The average result was taken as the final evaluation result to improve the reliability and generalization ability of the evaluation.

[0090] The classification model's performance is evaluated using classification accuracy, while the prediction model's performance is evaluated using the coefficient of determination (R²), mean squared cross-validation error (RMSEC), mean squared prediction error (RMSEP), and relative percentage deviation (RPD). Higher R² and RPD, and lower RMSEC and RMSEP, indicate a more reliable model. A RPD greater than or equal to 3.0 indicates significant predictive performance; between 2.5 and 3.0, the predictive performance is acceptable; and less than or equal to 2.5 requires further optimization. The main model evaluation metrics are listed above to ensure the comprehensiveness and accuracy of the model evaluation.

[0091] The classification fusion strategy employs feature-level fusion and decision-level fusion to comprehensively analyze data extracted from computer vision, electronic nose, and electronic tongue, thereby constructing a classification model. In the feature-level fusion strategy, color features (9 dimensions), aroma features (14 dimensions), and taste features (18 dimensions) representing instant green tea powder and tea liquor are first extracted from the raw information from computer vision, electronic nose, and electronic tongue. Next, the features of tea powder and tea liquor are merged pairwise to generate four new two-source fusion feature sets. Finally, SVM and RF classification models are constructed based on the fused feature datasets. Similar to the feature-level fusion strategy, the feature extraction process involves independently building corresponding SVM and RF classification models based on features from each data source to obtain their respective decision results. By integrating the model decision results from computer vision, electronic nose, and electronic tongue, these results are then input into an MLR model and DS evidence theory to achieve decision-level data fusion and ultimately generate the classification result.

[0092] The results of multi-feature fusion are shown in Tables 1-4 below:

[0093] Table 1. Classification results of instant green tea categories using SVM and RF based on feature-level fusion strategies.

[0094]

[0095] Table 2 Classification results of MLR models based on decision-level fusion strategy

[0096]

[0097] Table 3. Classification results of DS evidence theory based on decision-level fusion strategy

[0098]

[0099] Table 4. Prediction results of quality index content based on fused feature sets

[0100]

[0101] As shown in Tables 1-4, this embodiment organically combines three intelligent evaluation technologies—computer vision, electronic nose, and electronic tongue—to construct a multi-source data fusion platform. Computer vision technology, through image acquisition and processing, can accurately extract the color features of instant green tea powder and tea liquor. These features intuitively reflect key information such as the oxidation level and pigment distribution of instant green tea. Electronic nose technology simulates the olfactory system of mammals, enabling objective evaluation of tea aroma quality by detecting the aroma components of instant green tea. Electronic tongue technology further simulates the human taste system, accurately analyzing the taste characteristics of instant green tea and providing more comprehensive data support for quality evaluation.

[0102] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Those skilled in the art should understand that although the present invention has been described in detail with reference to the foregoing embodiments, modifications can still be made to the technical solutions of the foregoing embodiments, or equivalent substitutions can be made to some of the technical features, and these modifications or substitutions do not exceed the protection scope of the present invention, which is determined by the appended claims.

Claims

1. A multi-source data fusion-based instant green tea quality intelligent evaluation method, characterized in that, The intelligent evaluation method comprises the following steps: Step one: Selecting instant green tea as raw material, brewing a part of instant green tea with boiling water as tea soup, and taking another part of instant green tea as tea powder material, collecting visual images of the tea soup and the tea powder material by using a computer vision device, obtaining aroma information of the tea soup and the tea powder material by using an electronic nose, collecting taste information of the tea soup by using an electronic tongue, and measuring quality component information in the tea soup and the tea powder by using high performance liquid chromatography; Step two: Dividing the data set after dimension reduction processing of the aroma, visual and taste information as characteristic parameter data, determining quality indexes of the tea soup and the tea powder material, performing correlation analysis on the quality indexes and the characteristic parameters, and forming a fusion feature set; Step three: feature selection is performed on the fusion feature set formed in step two to screen out an optimal feature subset, based on which a regression classification model and a prediction model are respectively constructed to predict the quality index content of the instant green tea, and the classification accuracy, R 2 values of the training set and the prediction set, RMSEC, RMSEP and RPD indexes are used as evaluation criteria; Step four: Based on the optimal feature subset formed in step three, data fusion is used at the decision level and the feature level. In the model architecture at the decision level, multiple linear regression model and D-S evidence theory are used. In the model architecture at the feature level, feature selection method is introduced to perform multi-source fusion on the data. According to the evaluation results, the fusion model is obtained. For the variety identification of instant green tea, the model architecture at the decision level is used. For the quality prediction of instant green tea, the model architecture at the feature level is used. R 2 The values of RMSEC, RMSEP and RPD are used as evaluation indexes to select the best fusion model. 2.The multi-source data fusion based instant green tea quality intelligent evaluation method according to claim 1, characterized in that, The computer vision device comprises: a dark box, a support, a light source, a camera and a containing container; The support, the light source, the camera and the containing container are all arranged in the dark box; The light source and the camera are respectively connected with the support; The camera, the light source and the containing container are arranged in sequence from top to bottom; The containing container is used for containing the tea soup or the tea powder material. 3.The method of claim 2, wherein the method is characterized by, The dark box comprises a black matte acrylic plate in the shape of a cube, a side sliding groove and a bottom groove are arranged on the front side of the dark box to realize the sliding of the front panel up and down and the insertion into the bottom groove, so that a closed space is formed in the dark box to avoid the influence of external light environment on sample collection; The support comprises a bearing table, a supporting rod and an adjusting frame, the bearing table and the supporting rod are connected, the adjusting frame and the supporting rod are connected, a stepped circular groove is designed at the center point of the bearing table for fixing the containing container to ensure the consistency of the collection position each time, and the adjusting frame is used for adjusting the distance between the light source, the camera and the containing container; The light source adopts a ring-shaped shadowless light source; The camera is connected with a computer through a data line and the camera parameters are adjusted by using software to realize image collection. 4.The multi-source data fusion based instant green tea quality intelligent evaluation method according to claim 1, characterized in that, The olfactory sensing detection system comprises an electronic nose; The gustatory sensing system comprises an electronic tongue. 5.The multi-source data fusion based instant green tea quality intelligent evaluation method according to claim 1, characterized in that, The regression classification model comprises one or more of SVM, KNN and RF; The prediction model comprises one or more of PLSR, SVR and RF. 6.The multi-source data fusion based instant green tea quality intelligent evaluation method according to claim 1, characterized in that, The model evaluation indexes in the evaluation criteria and the calculation methods are as follows: Classification accuracy ; coefficient of determination R 2 values ; Cross-validated root mean squared error ; Root mean square error of prediction ; standard deviation ; Relative percent deviation ; where n is the number of samples in the dataset; X i is the actual value of the i-th sample in the process of establishing the prediction model; is the average of the actual values of all samples in the process of establishing the prediction model; Y i is the predicted value of the i-th sample in the process of establishing the prediction model; is the average of the predicted values of all samples in the process of establishing the prediction model. 7.The multi-source data fusion based instant green tea quality intelligent evaluation method according to claim 1, characterized in that, The determination of the quality indexes of the tea soup and the tea powder material comprises the following processes: determining tea polyphenols, total free amino acids, the ratio of tea polyphenols to total free amino acids, caffeine, epigallocatechin, catechin, epicatechin and epigallocatechin gallate by using liquid chromatography. 8.The method of claim 1, wherein the method is characterized by, In step two, PCA dimension reduction processing is adopted. 9.The multi-source data fusion based instant green tea quality intelligent evaluation method according to claim 1, characterized in that, The characteristic selection method comprises one or more of four characteristic selection methods, i.e. Pearson score, recursive feature elimination, particle swarm optimization and Lasso regression.

10. The application of the instant quality evaluation method of multi-source data fusion based instant green tea according to claims 1-9, characterized in that, The method is used for intelligent identification and prediction of instant green tea quality.

Images